Capturing Environmental Context in Real World Mobility: A Comparison between Machine Learning Based Indoor Outdoor Classification From Wearable Barometric Pressure and Magnetic Field

This Zenodo record contains raw barometric and magnetometer data from one representative subject of the independent test set, together with the corresponding window-level ground-truth labels and the trained machine-learning models supporting the study on indoor–outdoor discrimination during real-world mobility. The shared test subject corresponds to anonymized Subject ID 3. The data include raw barometric pressure signals, raw three-axis magnetometer signals, and indoor–outdoor ground-truth labels computed at window level. Labels are encoded as 1 for indoor and 0 for outdoor. The best model used for indoor–outdoor classification, trained with barometric data, is also provided. The study was approved by the Comitato Etico Territoriale Interaziendale AOU Maggiore della Carità di Novara, Italy, under Protocol No. 474/CE (2024).